fluff-cutter 0.4.0

A CLI tool to analyze academic papers and extract their core value

Paper Fluff Cutter

A CLI tool that cuts through academic paper fluff to extract what actually matters.

Most research has close to zero value. This tool uses multimodal LLMs to analyze papers and answer the three questions every paper should be able to answer:

1. Why should I care? - What problem does this address and why does it matter?
2. What's the actual innovation? - What's the core idea in plain terms?
3. Is the evidence convincing? - Do the experiments actually support the claims?

Installation

pip install fluff-cutter

Requires Python 3.10+.

Development install

git clone https://github.com/weijianzhg/paper-fluff-cutter.git
cd paper-fluff-cutter
pip install -e .

Configuration

Option 1: Interactive setup (recommended)

fluff-cutter init

This will prompt you for your API keys, default provider, and model preferences, then save them to ~/.fluff-cutter/config.yaml.

Option 2: Environment variables

export OPENAI_API_KEY=sk-your-key-here
export ANTHROPIC_API_KEY=sk-ant-your-key-here
export OPENROUTER_API_KEY=sk-or-your-key-here
export FLUFF_CUTTER_PROVIDER=anthropic  # optional, default provider
export FLUFF_CUTTER_OPENAI_MODEL=gpt-5.2  # optional, override default model
export FLUFF_CUTTER_ANTHROPIC_MODEL=claude-sonnet-4-5  # optional, override default model
export FLUFF_CUTTER_OPENROUTER_MODEL=anthropic/claude-sonnet-4-5  # optional, override default model

Usage

Basic usage

fluff-cutter analyze paper.pdf

By default, the analysis is saved to a .md file with the same name as the input (e.g., paper.pdf → paper.md).

Analyze from URL

You can pass a URL directly instead of a local file. The PDF will be downloaded to the current directory and then analyzed:

fluff-cutter analyze https://arxiv.org/pdf/2411.19870

Arxiv abstract URLs (/abs/) are automatically converted to PDF URLs. If the PDF has already been downloaded, it will be reused without re-downloading.

Specify output file

fluff-cutter analyze paper.pdf --output analysis.md

Print to stdout

fluff-cutter analyze paper.pdf --print

Specify provider

fluff-cutter analyze paper.pdf --provider openai
fluff-cutter analyze paper.pdf --provider anthropic
fluff-cutter analyze paper.pdf --provider openrouter

Specify model

fluff-cutter analyze paper.pdf --provider openai --model gpt-5.2
fluff-cutter analyze paper.pdf --provider anthropic --model claude-sonnet-4-5
fluff-cutter analyze paper.pdf --provider openrouter --model google/gemini-2.5-pro

Long papers

For very long papers that exceed the model's token limit, you can limit the number of pages:

fluff-cutter analyze paper.pdf --max-pages 30

If you don't specify --max-pages and the paper exceeds the token limit, it will automatically truncate to the first 50 pages and retry.

Supported Providers

Provider	Default Model	Environment Variable	Notes
OpenAI	gpt-5.2	OPENAI_API_KEY	Native PDF support
Anthropic	claude-sonnet-4-5	ANTHROPIC_API_KEY	Native PDF support
OpenRouter	anthropic/claude-sonnet-4-5	OPENROUTER_API_KEY	Access to 300+ models

All providers support PDF input natively - no external dependencies like poppler needed.

Configuration Precedence

Configuration is loaded with the following precedence (highest to lowest):

1. Command-line arguments (--provider, --model)
2. Environment variables (FLUFF_CUTTER_PROVIDER, FLUFF_CUTTER_*_MODEL)
3. Config file (~/.fluff-cutter/config.yaml)
4. Provider defaults

License

MIT